Signature redacted for privacy.

Transcription

1 AN ABSTRACT OF THE THESIS OF Marcia H. Wilson fcr the degree of Doctor of Philosophy in Fisheries and Wildlife presented on February 9, Title: Comparative Ecology of Bobwhite and Scaled Quail in Southern Texas Abstract approved: Signature redacted for privacy. John A. Qfawford For my research goals I addressed the theoretical importance of competition in habitat selection of 2 quails and examined the use of resources in peripheral and core areas. Methodological objectives were Co determine cover and food selection of Texas bobwhite (Colinus virginianus texanus) and chestnut-bellied scaled quail (Callipepla squamata castanogastris) in southern Texas. The Welder Refuge was the core area for bobwhites. The Chaparrosa and Experiment Ranches were peripheral areas for both quails. The core population for scaled quail was located at the Killam Ranch. Available cover formed a gradient from Welder to Killam. The highest percentage of herbaceous material was on Welder, whereas the largest amount of bare ground and shrub cover was at Killam. Bobwhites and scaled quail used and preferred different cover types. Bobwhites selected dense herbaceous material, whereas scaled quail preferred sparse vegetation and a shrub overstory. Based on niche

2 breadth estimates, both quails used a narrow range of cover types in their respective core areas and a broad range in the peripheral areas. The frequency of available food items also formed a gradient. Welder had the largest percent frequency of animal matter and of herbaceous seeds and green leaves. Killam had the highest frequency of seeds, leaves, and fleshy fruits from woody plants. Bobwhites preferred grass seeds and animal matter, whereas scaled quail selected the fruits and seeds of woody plants. Niche breadth estimates were relatively constant for each quail. Theoretically, when intraspecific competition is most intense a species selects a broad range of habitats, whereas sympatric species occupy a narrow range of resources due to interspecific competition. However, when I used niche breadth estimates to address my first goal, neither component of the habitat (cover or food) followed the anticipated pattern. Therefore, I believe competition is not the major selective force in habitat selection of quails. The generalization has been put forth that in the peripheral part of a range birds occupy their optimal habitat, whereas at the center of the range, optimal and suboptimal habitats are used. However, the core area contains an abundance of preferred habitat and a species is able to meet its life-history requirements whithin a narrow range of cover types. Peripheral areas supply only suboptimal habitat and in order to fulfill the species-specific needs a species must use a wide range of cover types.

3 Comparative Ecology of Bobwhite and Scaled Quail in Southern Texas by Marcia H. Wilson A THESIS submitted to Oregon State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Completed February 9, 1984 Commencement June 1984

4 APPROVED: Signature redacted for privacy. Associ t Professor of Wi dlife ology in charge of major Signature redacted for privacy. Head of Department of Fisheries and Wildlife Signature redacted for privacy. of Graduat3 (Dean ctiool Date thesis is presented February 9, 1984 Typed by Kelly D. Schmidt for Marcia H. Wilson

5 ACKNOWLEDGEMENTS Financial support for this study was received from the Rob and Bessie Welder Wildlife Foundation. I appreciated the assistance of Dr. James Teer, Director, and Dr. Lynn Drawe, Assistant Director. I am grateful to my committee chairman, Dr. John Crawford, for his guidance and patience. Likewise, I want to thank the members of my committee, Dr. Bill Krueger, Dr. Charles Warren, Dr. Gordon Matzge, and Dr. Dick Clinton for all their time. Many people made my work plausible on the study areas. At the Chaparrosa Ranch, Malcolm Richey was always willing to lend a hand if any difficulty arose. Dr. Pat Reardon provided background information. Dr. Lytle Blankenship was very helpful concerning the quail on the Experiment Ranch. For the Killam Ranch, Frank Matthew's advice and assistance was greatly appreciated. the following friends helped me collect quail: At the Welder Refuge Mike Passmore, Larry Williams, Bruce Thompson, and Deb Hoffmaster. A very special thanks must be extended to my husband, Andy, for his time and companionship while in the field, his constant maintenance of the VW, and his encouragement throughout the course of this study.

6 TABLE OF CONTENTS Pase INTRODUCTION 1 RESEARCH AREAS 5 METHODS 10 Sampling Methods for Cover 10 Sampling Methods for Food 11 Selection and Niche Breadth Indices 12 Statistical Analysis 13 RESULTS AND DISCUSSION 16 Quail Population Analysis 16 Indices to Abundance 16 Cover Analysis 16 Annual Cover Availability Among Study Areas 16 Seasonal Cover Availability on All Study Areas Combined 20 Annual Cover Used by Quail Among Study Areas 20 Annual Cover Used by Quail on All Study Areas Combined 23 Seasonal Cover Used by Quail on All Study Areas Combined 25 Selection Indices for Cover Among Study Areas 27 Selection Indices for Cover on All Study Areas Combined 29 Niche Breadth Indices for Cover Among Study Areas 29 Niche Breadth Indices for Cover on All Study Areas Combined 32 Food Analysis 34 Annual Food Availability Among Study Areas 34 Seasonal Food Availability on All Study Areas Combined 36 Annual Food Habits Among Study Areas 36 Annual Food Habits on All Study Areas Combined 41 Seasonal Food Habits on All Study Areas Combined 44 Selection Indices for Food Among Study Areas 47 Selection Indices for Food on All Study Areas Combined 50 Niche Breadth Indices for Food Among Study Areas 50 Niche Breadth Indices for Food on All Study Areas Combined 54 CONCLUSION 55 LITERATURE CITED 59 APPENDICES 64

7 LIST OF FIGURES Figure yase Distribution of bobwhite and scaled quail throughout North America. 2 Distribution of Texas bobwhites (vertical) and chestnut-bellied scaled quail (horizontal) in relation to the location of study areas Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Percent cover used by bobwhites (BW) and scaled quail (SCQ) on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Cover selection indices for bobwhites (BW) and scaled quail (SCQ) on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas,

8 LIST OF TABLES Table Pase Indices to abundance for bobwhite (BW) and scaled (SCQ) on 4 study areas, Texas, Relative percent of available cover by cover type height categories on the Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Relative percent of available cover separated by breeding and nonbreeding season on all 4 study areas combined, Texas, Percent cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Percent cover used by bobwhites (BW) and scaled quail (SCQ) during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Selection indices for cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Selection indices for cover used by bobwhite (BW) and scaled quail (SCQ) during the breeding and nonbreeding seasons on all 4 study areas combined, Texas,

9 Table Page Niche breadth indices for cover (grass, forb and bare ground) used by bobwhites and scaled quail by each study area, on all 4 study areas combined, and by season, Texas, Percent frequency of available food items on the Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Percent frequency of available food items during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Relative percent of diet for bobwhites (BW) and scaled quail (SCQ) on the Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches and by season, Texas, Relative percent of diet for bobwhites (BW) and scaled quail (SCQ) on the Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Relative percent volume of annual diet for bobwhites (BW) (N = 120) and scaled quail (SCQ) (N = 88) on all 4 study areas combined, Texas, Percent volume of diet for bobwhites (N=120) and scaled quail (N=88) on all 4 study areas combined, Texas,

10 Table Page Relative percent of diet for bobwhites and scaled quail during the breeding (B) and nonbreeding (NB) season on all 4 study areas combined, Texas, Selection indices for food items by bobwhites (BW) and scaled quail (SCQ) on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Selection indices for food items by bobwhites and scaled quail on all 4 study areas combined, Texas, Selection indices for food items by bobwhites and scaled quail during the breeding (B) and nonbreeding (NB) season on all 4 study areas combined, Texas, Niche breadth indices for the diet (seed, leaf, fruit, flower, blub, animal) of bobwhite and scaled quail by each study area, on all 4 study areas combined, and by season, Texas,

11 APPENDIX CONTENTS Figures Page Scatterplot of the first and second functions derived from discriminant analysis for the percent cover of grass, forb, grass/forb, woody and bare ground available on Welder (1), Chaparrosa (2), Experiment (3), and Killam (4) Ranches, Texas, Scatterplot of the first and second functions derived from discriminant analysis for the percent cover of grass and forb height categories plus the lifeforms of woody cover available on Welder (1), Chaparrosa (2), Experiment (3), and Killam (4) Ranches, Texas, Scatterplot of the first and second functions derived from discriminant analysis for the percent frequency of seeds, leaves, fruits, flowers, bulbs, and animal matter on Welder (1), Chaparrosa (2), Experiment (3), and Killam (4) Ranches, Texas Scatterplot of the first and second functions derived from discriminant analysis for the partitioned seeds, leaves, and animal matter categories on Welder (1), Chaparrosa (2), Experiment (3), and Killam (4) Ranches, Texas,

12 Tables Page Seasonal rainfall for Welder (W), Experiment (E), and Killam (K) Ranches, Texas, (Data for Chaparrosa were available only during breeding season.) 68 Classification results for available vegetation on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Pooled-within-group correlations between the discriminant functions (DF) and discriminating variables for the available vegetation on Welder, Chaparrosa, Experiment, and Killam Ranches, Texas, Pooled within-group correlations between discriminant function and discriminating variables for the available vegetation and during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Classification results for cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Pooled within-group correlations between discriminant function and discriminating variables for cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Pooled within-group correlation between the discriminant function and discriminating variables for cover used by bobwhites and scaled quail during the breeding season on all 4 study areas combined, Texas,

13 Tables Page Pooled within-group correlations between the discriminant function and discriminating variables for cover used by bobwhites and scaled quail during the nonbreeding season on all 4 study areas combined, Texas, Pooled within-group correlations between the discriminant function and discriminating variables for cover used by bobwhites during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Pooled within-group correlations between the discriminant function and discriminating variables for cover used by scaled quail during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Classification results for available food items on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, Pooled-within-group correlations between discriminant functions (DF) and discriminating variables for available food items on Welder, Chaparrosa, Experiment, and Killam Ranches, Texas, Pooled within-group correlations between the discriminant function and discriminating variables for available food items during breeding and nonbreeding seasons on all 4 study areas combined, Texas,

14 Table Page N. Pooled within-group correlations between discriminant function and discriminating variables for food items consumed by bobwhites and scaled quail on all 4 study areas combined, Texas, Pooled within-group correlations between discriminant function and discriminating variables for food items consumed by bobwhites and scaled quail during the breeding season on all 4 study areas combined, Texas, Pooled within-group correlations between discriminant function and discriminating variables for food items consumed by bobwhites and scaled quail during the nonbreeding season on all 4 study areas combined, Texas, Pooled within-group correlations between discriminant function and discriminating variables for food items consumed by bobwhites during the breeding and nonbreeding season on all 4 study areas combined, Texas, Classification results for food items consumed by bobwhites (BW) and scaled quail (SCQ) during 2 seasons and throughout the year on all 4 study areas combined, Texas,

15 COMPARATIVE ECOLOGY OF BOBWHITE AND SCALED QUAIL IN SOUTHERN TEXAS INTRODUCTION The North American aridlands apparently developed from the time of the Pliocene or earlier (Hubbard 1973). During Pleistocene glaciation, the aridlands became fragmented. At this time widespread ancestors for the genera Callipepla and Colinus could have become disjoined into isolated refugia (Hubbard 1973). Scaled quail (Callipepla squamata) probably became differentiated as a separate species in the Chihuahuan Refugium, whereas the bobwhite (Colinus virginianus) became differentiated in the Flordia Refugium (Hubbard 1973). After glaciation a more continuous distribution of aridlands returned. The scaled quail expanded its distribution to encompass the present-day Chihuahuan Desert and the bobwhite extended its range westward and northward to occupy woodland and/or brush-grassland habitats (Johnsgard 1973). These 2 quails exhibit a certain amount of geographic overlap along the edge of their ranges (Fig. 1). Along this edge, Jackson (1942), Schemnitz (1964), and Hatch (1975), found that scaled quail (C. s. pallida) and bobwhites (C. V. taylori) generally occupied different cover types within an area. In Arizona, Anderson (1974) found scaled quail and Gambel's quail (Callipepla gambelii) in different habitats. Likewise, Gutierrez (1977) found that California quail (Callipepla californicus) and mountain quail (Oreortyx pictus) used species-specific components of the vegetation in central California. These coexisting quails were on the margins of their distributions. According to Hilden (1965), bird

16 r*: a. BOBWHITE SCALED QUAIL OVERLAP Fig. 1. Distribution of bobwhite and scaled quail throughout North America.

17 3 species are stenotopic at the periphery of their distribution and more eurytopic in the center of their breeding range. The reason Hilden (1965) made this generalization was that the range of habitat (cover, food, water, and space) exploited by a species was dependent on the density of a population. If the density were low such as on the periphery of the distribution birds occupied only the optimal habitat; with increased density less suitable habitat was also occupied. Not only do the areas of overlap contain populations of quail on the periphery of their range, these areas also accommodate coexisting quails that are taxonomically related (Holman 1961) and ecologically similar. According to Schoener (1982), interspecific competition is the major selective force acting on similar sympatric species. Interspecific competition results in the partitioning of the habitat wherein each species uses a narrow range of resources (Diamond 1978). Schoener (1974) found that a preponderance of studies indicated that, cover, primarily, and food, secondarily, were the prevalent factors in habitat partitioning. Away from the area of overlap Svardson (1949) and Hilden (1965) proposed that each species uses a broad range of resources because of intraspecific competition. These authors seemed to understand competition as nonconditional on other factors. In contrast, other ecologists understand competition more as part of a complex of interactions. Thus, the importance of competition may be flexible and/or contingent on a variety of interrelated elements. For example, Wiens (1977) warned that the different patterns of resource use among or between coexisting species may be explained by an intermittency of competition or by other

18 4 processes. Thomson (1980) maintained that mutualism or a combination of interspecific interactions may be involved in changes in population numbers and niche shifts. Connell (1975) reported that predation might lower populations below the point at which competition occurred. Based on a literature review, Jackson (1981) reported that plant ecologists, since the early 1900s, have been interested in the combined effects of competition, predation, and physical environmental disturbances on plant community composition. Hamilton (1962) raised the issue that in order to attribute resource partitioning to the avoidance of competition, habitat requirements outside the area of sympatry must be assessed. Yet few, if any studies, have been conducted concurrently on sympatric and allopatric populations. The goals of this study were to address the possible theoretical influence of competition on cover and food partitioning between 2 quails and to examine the use of habitats in peripheral and core (highest densities of 1 species) areas. More specifically, my methodological objectives were to determine cover and food selection of Texas bobwhite (Colinus virginianus texanus) and chestnut-bellied scaled quail (Callipepla squamata castanogastris) in core (allopatric) areas as well as peripheral (overlap) areas.

19 5 RESEARCH AREAS I selected 4 study areas in southern Texas (Fig. 2). The allopatric/core area for bobwhites was located at the Rob and Bessie Welder Wildlife Refuge. The refuge is situated in a transitional zone between the Gulf Prairies and Marshes and the South Texas Plains (Thomas 1975). The study areas with coexisting populations of Texas bobwhites and chestnut-bellied scaled quail were located within the South Texas Plains (Thomas 1975). The peripheral sites for both species were on the Chaparrosa Ranch and the Rio Grande Plain Experiment Ranch. The Killam Ranch was the core area for the scaled quail, as well as a peripheral site for bobwhites. All the study areas were operating cattle ranches. The Welder Refuge, located approximately 80 km northeast of Corpus Christi, occupies 3,157 ha and borders the Aransas River in San Patricio County. The annual rainfall from 1956 to 1977 was 89 cm with extreme fluctuations between years (38 cm in 1956 and 125 cm in 1973) (Drawe et al. 1978). The growing season exceeds 300 days per year with vegetative growth peaks in the spring and fall. characterized by hot, humid summers and mild winters. The area is The evaporation rate is 150 cm (Moore, pers. comm.). On the uplands, the slope gradient is less than 1%. Of the 12 range sites, the most extensive site is the Blackland (Victoria clay soil), which is level and drains poorly. Shrub-grassland dominates the uplands; the predominant shrubs include mesquite (Prosopis glandulosa) and blackbrush (Acacia rigidula). The common mid-grasses are seacoast bluestem (Scizachyrium

20 6 III W Atall IIIM einiiiiiii 111 MEM MEW INIOallIIIIIIIIIIW M# II I II MMOMM NIMWM1111MINIMIIINIMINI. VEMEMMOMMIOMMIIMMEML 11111MMIMOSMIMIli M ' M11 I IIMEMLIMOMMEMMUMM 'MIMI Ill II III MINI IN II WM M / i 'MOMMEMMEM 1r I` Fig. 2 Distribution of Texas bobwhites (vertical) and chestnutbellied scaled quail (horizontal) in relation to the location of study areas Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas,

21 7 littoralis), meadow dropseed (Sporobolas asper), and Texas wintergrass (Stipa leucotricha). After 1975, a year-round grazing system was replaced by several specialized systems. In order to control woody plants, most of the refuge has been either roller chopped, rootplowed, or sprayed with kerosene (Drawe et al. 1978). The Chaparrosa Ranch is approximately 11 km west of La Pryor in Zavala County. Both the East and West Big Bowles pastures were selected as study sites. These pastures totaled 2,467 ha. For Zavala county the average annual precipitation is 55 cm (Escobar, pers. comm.). As in other research areas in the South Texas Plains, the Chaparrosa has the highest rainfall in May and June. This period is typically followed by a summer drought with another peak in precipitation during September (Shaw and Dodd 1976). rate for the county is 183 cm (Escobar, pers. comm.). The evaporation The topography for the ranch is primarily flat with slopes ranging from 0 to 8%. Of the 9 range sites, the most predominant are the Clay Loam (58%) and the Clay Flat (11%). Only about 2% of the pastures had Shallow Ridge and Rolling Hardland range sites, and these latter sites were around a caliche pit. The vegetation on the clay sites was characterized by the grass, curly mesquite (Hilaria belangeri), and the annual forb, bundlef lower (Desmanthus sp.). The woody cover consisted of mesquite and whitebrush (Aloysia lyciodes). The Rolling Hardland site was comprised of guajillo (Acacia berlanderi) with an understory of short, sparse grasses such as threeawn (Aristida 22.) and annual forbs. Major portions of the Bowles pastures were sprayed with the herbicide 2,4,5-T between 1971 and 1976.

22 8 The Rio Grande Plain Experiment Ranch is located about 45 km southwest of Uvalde in northeastern Maverick and southeastern Kinney counties. This 4,228 ha ranch is operated by the Texas Agricultural Experiment Station. My research was conducted on pastures numbered 1 through 13. For Maverick County, rainfall from 1936 to 1968 averaged 50 cm (Clemente, pers. comm.) and ranged from 13 cm to 113 cm. The average growing season was 285 days. Nine range sites comprised the research area. As at the Chaparrosa Ranch, the most extensive site were the Clay Loam (58%) and the Clay Flat (10%). The Rolling Hardland and the Shallow Ridge sites made up about 7% of the area. Three specialized grazing systems were in use at the ranch. Parts of pastures 1, 2, 3, 8, and 9 were tame pastures and seeded with buffelgrass (Cenchrus ciliaris), an introduced warmseason bunchgrass. Other small sections of the area were chopped to control dense brush. The Killam Ranch surrounds Laredo in Webb County. My research was restricted to the North and South Charco Pastures as well as a portion of the Middle Pasture, a total of about 4,000 ha. The study sites were located about 18 km north of Laredo. The annual precipitation for Webb County is 50 cm (Nelle, pers. comm.). The growing season exceeds 300 days annually and the average evaporation is 193 cm (Nelle, pers. comm.). Seven range sites were found on the 3 pastures. Unlike the Chaparrosa and Experiment Ranches, only 7% of the area was typified by Clay Loam and Saline Clay range sites, whereas over 40% of the area was Rolling Hardland and Gravelly Ridge sites. For the Gravelly Ridge sites the gradients averaged less than 10%. Guajillo and blackbrush are the dominant woody plants with

23 9 threeawn and Texas bristle grass (Setaria texana) as the grass understory. Prickly pear cactus (Opuntia 211.) is more predominant on the Killam Ranch than any of the other research areas. The grazing system involved a 6-month rotation between the North and South Charco. On the western border of the South Charco periodic fires occurred along the railroad tracks. The Middle Pasture was chained several years ago. Portions of the Charco Pastures were rootplowed and were dominated by huisache trees (Acacia farnesiana), prickly pear cactus, and buffelgrass.

24 10 METHODS Field research was conducted for 1 annual cycle from April 1980 through March of The cycle was subdivided into breeding (April through September) and nonbreeding seasons (October through March). Female bobwhites have been collected with eggs in the oviduct every month of the year (Cain, pers. comm.), but a large majority of breeding takes place from April through September. Both cover and dietary data were collected 6 times throughout the year on each of the 4 study areas. Sampling Methods for Cover To characterize the available cover on each research area I used 100 random points. I conducted a 1-hour flush census in both the morning and evening with the aid of a bird dog. In order to determine the cover used by quail, the initial sightings of quail such as pairs, calling males, broods, nests, roosts or coveys were marked with flags. Each random point (available cover) and sighting (cover used by quail) served as the locus of a 0.04 ha circular plot (James and Shugart 1970, James 1971). At each locus a 25-m line transect was established in a random direction. The line-intercept method was used to determine percent cover (Canfield 1941). Presumably, birds have species-specific preferences for visual structures of the vegetation. Pitelka (1941) found no consistent relationship between the dominant plants and bird species, but he did find a constant relationship between the distribution of birds and

25 11 lifeform of the vegetation. Likewise, Hamilton (1962) found that avian habitats in the Mesquite Plains of Texas reflected the structure or physiognomy of dominant vegetation, not the species composition. Thus, I used Durietz's lifeform classification as modified by Jones (1963) to characterize the vegetation of each circular plot. The lifeform categories included grass, forb, shrub, tree and cactus. The herbaceous cover was placed into height classes: under 5 cm, 10 cm, 20 cm, 40 cm, 80 cm, or over 80 cm. The 25-m transect was divided into 5-cm intervals. For each interval if 50% or more was intercepted by a grass or forb category, it was recorded. When both grass and forb occupied an interval, this double stratum was recorded and used as an index of dense herbaceous material, otherwise, bare ground was recorded. Furthermore, I recorded the woody plants (shrub, tree, cactus) if they intercepted 50% or more of an interval. Sampling Methods for Food In order to avoid a common shortcoming of many food studies (Gullion 1966), dietary information was collected during each sampling period. I tried to collect 1 quail from a covey or pair because quail within a social unit tend to have similar contents in their crops (Frye 1954). The availability of food items (percent frequency) was determined at the location where a quail was collected. After the utilized cover data were recorded, 10 ground samples were taken at equal intervals along the 25-m line. Within a 10 X 10-cm quadrat, all the ground

26 12 litter to an average depth of 2 cm was collected (Glading et al. 1940). These samples provided information about the frequency of grass, forb, and woody seeds as well as the frequency of snails (Gastropoda), onion bulbs (Allium a.), and termites (Isoptera). Twenty-five 10 X 10-cm quadrats were established along the transect to determine the frequency of green leaf material (grass, forb, woody), flowers, fruits, and seedheads. Also along the transect, animal matter was sampled with a sweep net at 25 stops. Food habits were determined by examination of the crop contents. If the number of crops collected during 1 of the 6 sampling periods amounted to more than a third of the total sample of crops from 1 study area, I averaged the number of crops taken for the 5 periods. I randomly selected crops from the over-represented period to equal the average determined from the rest of the sampling periods. I froze the crops to facilitate identification of material at a later date. Food items were separated into the following categories: seeds (grass, forb or woody), leaves (grass, forb or woody), flowers (forb or woody), fleshy fruits, bulbs and animal matter (spiders (Arachnida), snails and insects). I measured the volume of each item by water displacement to the nearest 0.02 cc. Frequency of food categories was also recorded. Selection and Niche Breadth Indices The proportion of available resources used by each species provided an index of habitat selection. For cover selection I used Ivlev's (1961) index of electivity E = r + p' r is equal to the

27 13 proportion of cover used and p is equivalent to the portion of available cover. For this index, -1 indicated avoidance and +1 suggested selection for the particular cover types. Dietary selection % frequency X % volume was estimated by the following formula: SI = % availability This index is similar to the desirability coefficient proposed by Glading et al. (1940). Availability of animal matter (except for snails and termites) was collected differently than the leaves (25 quadrats) or the seeds (10 quadrats). Therefore, selection indices among these major sampling groups could not be compared. Pianka (1973) suggested an index that estimated the variability of resources (cover and food) used by each species. This estimate is commonly referred to as niche breadth (B). The equation, B = ' 7 isdeterminedforeachspeciesand pi proportion of a cover type or food item. is the relative Statistical Analysis To examine the relationships between groups (2 species, 4 locations or 2 seasons) and the simultaneous variation of all the variables (cover or diet), I used multivariate analysis (Sokol and Rohlf 1981). Each transect (cover used by quail) was recorded as either bobwhite or scaled quail regardless of the number of birds that were sighted at the location. Both data sets (cover and food) contained totals that were the sums of either height categories or plant parts. Because the totals formed singular matrices, I subdivided the data. The cover variables were separated into 2 segments: 1) the total percent of grass, forb, grass/forb, bare

28 14 ground and woody cover, 2) the height categories for grass and forb and the tree, shrub, and cactus classes. The food variables were divided into 3 parts: bulbs and animal matter 1) totals for seeds, leaves, fruits, flowers, 2) seeds (grass, forb, and woody), leaves (grass, forb, and woody), spiders, snails, and insects 3) compilation of plant parts into total grass, forb or woody categories and the total animal matter. For each division of the data, multivariate analysis of variance was performed to determine if the vectors of the means (centroids) were different between or among the groupings. Hotelling's Tsquare was used to test for differences among the 4 study areas (Nie et al. 1975). If the vectors were different at the 0.05 level (or less) then each segment was subjected to discriminant analysis. I used this type of analysis as a descriptive tool. Discriminant analysis combined the variables in a linear function, which maximized the separation of the groups (Nie et al. 1975). The discriminant functions can be visualized as axes in a geometric space and thus, they provide spatial relationships between or among the groups (Gauch 1982). The number of discriminant functions is 1 less than the number of groups. Interpretation of the function(s) was based on the correlation of each variable with the discriminant scores derived for each function (Raphael 1981). Variables with the highest correlations were used to interpret the functions (Raphael 1981). To check the adequacy of the discriminant functions a classification of the original set of cases within each group was conducted (Nie et al. 1975). Thus, a certain

29 15 percentage of cases were correctly classified by the discriminating variables I selected.

30 16 RESULTS AND DISCUSSION Quail Population Analysis Indices to Abundance Censuses provided indices to abundance (birds or quail sightings per hour) on each study area. I believe the scaled quail were insufficiently represented by the index of birds/hour because of the different escape behavior used by the quails. Coveys of bobwhites usually flushed as a unit, which provided a relatively accurate count. In contrast, a scaled quail covey generally separated and ran, offering glimpses of only a few birds. Therefore, I recorded coveys, pairs, or singles as 1 sighting regardless of the number of quail in the social unit, and used these sightings to estimate the ratio of bobwhites to scaled quail (Table 1). Both indices indicated that the Welder Refuge and Killam Ranch were the allopatric and/or core areas for bobwhites and scaled quail, respectively (Table 1). of total quail (Table 1), The Experiment Ranch had the lowest abundance possibly a result of the exceptionally dry nonbreeding season (Appendix, Table A). Apparently, the amount of rainfall is important to bobwhite populations at least (Lehman 1946). Cover Analysis Annual Cover Availability Among Study Areas The available cover types on the areas formed a gradient; Welder at 1 end had the most herbaceous cover and the least amount of bare

31 Table 1. Indices to abundance for bobwhite (BW) and scaled quail (SCQ) on 4 study areas, Texas, Birds/hr Sightings/hr Study area BW SCQ Total BW:SCQ BW SCQ Total BW:SCQ Welder : :00 Chaparrosa : :30 Experiment : :56 Killam : :86

32 18 ground and woody cover, whereas Killam had the smallest amount of herbaceous material but the largest amount of bare ground and woody cover (Table 2). The height classes of herbaceous vegetation also differed among areas. Welder was the area with the greatest amount of mid-level (< 20 cm, < 40 cm) and tall - (> 40 cm) herbaceous material. Killam possessed the least amount of mid- and tall herbaceous material (Table 2). In addition, the lifeforms of the woody material formed a gradient. Welder had the largest amount of tree cover, whereas Killam had the greatest amount of shrub and cactus cover (Table 2). At the 2 ends of the gradient, the core areas were significantly different from each other and from the other areas (Hotelling's T-square, P < 0.05). Welder was distinctive among all the areas (Appendix, Fig. A, B, and Table B) primarily because of the large percentage of total grass (817.), forb (28%), and grass/forb (217.) (Appendix, Table C). Within these major cover types the mid-grasses and forbs under 40 cm best characterized the study area with allopatric populations of bobwhites (Appendix, Table C). Total grass cover was highly correlated with grass under 20 cm (r = 0.70, P < 0.05) and with grass under 40 cm (r = 0.69, P < 0.05). These results supported the placement of Welder in a transitional zone between the South Texas Plains and the Coastal Prairies and Marshes. Within the South Texas Plains, the Killam Ranch was the most distinct area. The cactus and dense shrub cover characterized the core area for scaled quail (Killam Ranch) (Appendix, Figure B and Table C). Despite the smaller amount of forb cover (13%) on the Experiment

33 19 Table 2. Relative percent of available cover by cover type and height categories on the Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas, (N = 100 transects for each study area.) Percent cover by study area Cover type Grass (total): ShoEta Mid Tallc Forb (total): ShoEta Mid Tallc Grass/forb Bare ground Woody (total): Tree Shrub Cactus ashort = height categories under 5 cm and under 10cm, bmid = height categories under 20 cm and under 40 cm. ctall = height categories under 80 cm and over 80 cm.

34 20 Ranch, the available vegetation was not significantly different from the Chaparrosa, which had 20% forb cover (P < 0.05). Apparently, the Experiment Ranch was better described by height classes of herbaceous material than by the total percent of cover types (Appendix, Table C). Seasonal Cover Availability on All Study Areas Combined I contrasted the breeding and nonbreeding seasons by consolidation of available cover on all 4 study areas. The seasons were different (P < 0.05) primarily because of the increase in forb cover. More specifically, the short forbs increased from 3% in the breeding season to 15% in the nonbreeding season (Appendix, Table D). Otherwise, the composition of the cover types remained relatively constant from season to season (Table 3). This consistency probably was the result of the heavy rainfall in August (Hurricane Allen) and a second vegetative growth peak (Appendix, Table A). Annual Cover Used a Quail Among Study Areas Examination of cover types used by quail at each study area indicated that the core areas for bobwhites (Welder) and for scaled quail (Killam) were the most different from each other in terms of all the cover classes (Fig. 3). Secondly, all of the bobwhite populations (Welder through Killam) used different percentages of grass, bare ground and woody cover than did scaled quail population (Fig. 3). The exceptions were forb cover and grass/forb cover types; forb cover was strongly correlated with grass/forb (r = 0.80, P < 0.05).

35 21 Table 3. Relative percent of available cover separated by breeding and nonbreeding season on all 4 study areas combined, Texas, (N = 200 for each season.) Percent cover by season Cover types Breeding Nonbreeding Grass (total): Shea Mid Tallc 6 6 Forb (total): ShoEta 3 15 Mid 6 6 Tallc 3 2 Grass/forb 7 9 Bare ground Woody (total): Tree 9 5 Shrub Cactus 1 1 ashort = height-categories under 5 cm and under 10cm. bmid = height categories under 20 cm and under 40 cm. ctall = height categories under 80 cm and over 80 cm.

36 GRASS FORB % COVER USED BY QUAIL i E I K E C MMM Ei BW I SCQ M r,i i i I I I 1. K E GRASS KE C / FORB PḲ E C Igh C h C W w wc E K m n Y 9 I BARE C E K GROUND WOODY C K W C EK I r 0 n M vp I I I I-. 1 I, Fig. 3. Percent cover used by bobwhites (BW) and scaled quail (SCQ) on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas,

37 23 Nevertheless, at each study area bobwhites occupied areas with more forb and grass/forb cover than did scaled quail (Fig. 3). Annual Cover Used Quail on all Study Areas Combined I combined all the study areas and found that bobwhites used significantly different cover from scaled quail (P < 0.05). In addition, these 2 species were easily separated on the basis of cover types and height classes (Appendix, Table E). Bobwhites used denser herbaceous material, whereas scaled quail occupied areas with more bare ground and an overstory of woody cover (Table 4 and Appendix, Table F). Bare ground and total grass cover were negatively correlated (r = -0.76, P < 0.05). In addition, bobwhites were found in areas with taller herbaceous material, in particular mid-grasses, than were scaled quail (Appendix, Table F). Likewise, Schemnitz (1964) found in Oklahoma that scaled quail used vegetation that provided overhead- protection, but was open underneath. riparian areas with denser vegetation at ground level. Bobwhites used These differences in the profile of the cover probably reflected the different escape behaviors of the 2 quails; scaled quail run and bobwhites hide or "freeze" in dense herbaceous vegetation. Scaled quail used more shrub (37%) and cactus cover (3%) than did bobwhites (7% and 0%, respectively) (Appendix, Table F). Bobwhites were found more often in conjunction with tree cover than were scaled quail (Table 4). Similarly, Reid (1977) found that bobwhites in the South Texas Plain preferred mesquite habitats and deciduous woodlands, whereas scaled quail selected scrubland and brush habitats.

38 24 Table 4. Percent cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Percent cover Cover type Bobwhites Scaled quail (N = 466) (N = 292) Grass (total): ShoEta 20 9 Mid 32 8 Tallc 6 0 Forb (total): ShoEta 14 9 Mid 10 3 Tall' 4 0 Grass/forb 13 1 Bare ground Woody (total): Tree 19 4 Shrub 7 37 Cactus 0 3 ashort = height categories under 5 cm and under 10cm. bmid = height categories under 20 cm and under 40 cm. ctall = height categories under 80 cm and over 80 cm.

39 25 Furthermore, Lehman and Ward (1941) reported that prickly pear cactus was important to scaled quail in southern Texas. Seasonal Cover Used a Quail on All Study Areas Combined During the breeding season, bobwhites and scaled quail used different cover (P < 0.05). Bobwhites occupied areas with more total grass cover, especially mid-grasses, whereas scaled quail were associated with bare ground and shrub cover (Table 5 and Appendix, Table G). Lehman (1946) reported that in southwestern Texas during 1942 and 1943, 3 genera of grasses (Andropogon, Aristida and Manisuris) were the principal materials in 189 bobwhite nests, which amounted to 98% of the total he studied. In addition, during the 2 breeding seasons, the average heights of the grass around the nests were 23 cm and 30 cm (Lehman 1946). In contrast, Schemnitz (1964) reported only 3 of 50 scaled quail nests were located in grass. Information on the nests of chestnut-bellied scaled quail is minimal. I found 2 nests on the Experiment Ranch, both were located under shrubs. During the nonbreeding season both quails again used different cover (P < 0.05). Bobwhites were associated with denser herbaceous cover and scaled quail used areas with more bare ground and shrub cover (Appendix, Table H). For each quail, the cover used was compared between breeding and nonbreeding season. During the nonbreeding season both quails occupied areas with more forb cover, especially forbs under 5 cm, than they did during the breeding season (Appendix, Tables I and J). Bobwhites used denser vegetation during the nonbreeding season (19% to

40 26 Table 5. Percent cover used by bobwhites (BW) and scaled quail (SCQ) during breeding and nonbreeding seasons on all 4 study areas combined, Texas, Percent cover by season Breeding Nonbreeding BW SCQ BW SCQ Cover type (N = 233) (N = 162) (N = 233) (N = 130) Grass (total): ShoEta Mid Tallc Forb (total): ShoEta Mid Tallc Grass/forb Bare ground Woody (total): Tree Shrub Cactus ashort = height categories under 5 cm and under 10cm. bmid = height categories under 20 cm and under 40 cm. ctall = height categories under 80 cm and over 80 cm.

41 27 8%) and more bare ground in the breeding season (33% to 22%) (Appendix, Table I). for thermoregulation. Denser vegetation may have provided better cover Despite the minimal use of grass cover overall, scaled quail used more grass cover, in particular grass under 20 cm, during the breeding season (19%) (Appendix, Table J). Often, I found scaled quail in drainages during the breeding season, which may have reflected the dry conditions of other habitats and/or the need for brooding habitat. During the nonbreeding season scaled quail occupied denser brush than during the remainder of the year (Table 5 and Appendix, Table J). Selection Indices for Cover Among Study Areas All the bobwhite populations and the scaled quail populations had separate groupings of the selection indices for every cover type (Fig. 4). The coincidence of forb cover used by bobwhites at Killam and scaled quail at Chaparrosa (Fig. 3) was separated by accounting for the larger availability of forbs at the Chaparrosa (20%) than at Killam (10%). For bobwhites, the selection for total grass cover increased from Welder to Killam (Fig. 4). In fact, the bobwhite population at Welder showed a slight avoidance of grass cover and selection for bare areas (Fig. 4). Possibly, the quantity (81%) and rankness of grass at Welder inhibited movement for bobwhites, whereas the bare areas (12% of available cover) provided a more desired surface for walking, feeding, and roosting (Scott and Klimstra 1954, Klimstra and Ziccardi 1963). By contrast, scaled quail increased the amount of avoidance

42 SELECTION INDICES FOR COVER K GRASS III EC W FORB CE K 1ft11 n EK C I WK fai BW I SCQ CE GRASS K / FORB I BARE I I II i C I I E GROUND CKE 1 KC C E rq 1 q 1 M K I. I I I I I I I WOODY WKC E CE K I e i 1 rf)cm, I I I I I Fig. 4. Cover selection indices for bobwhites (BW) and scaled quail (SCQ) on Welder (W), Chaparrosa (C), Experiment (E), and Killam (K) Ranches, Texas,

43 29 for grass cover and increased their selection for woody cover from Chaparrosa to Killam (Fig. 4). Thus, shrubs may have increasingly replaced the use of grass as overhead cover. Selection Indices for Cover on All Study Areas Combined Comparison of selection indices annually and seasonally indicated that bobwhites and scaled quail preferred very different cover types (Tables 6 and 7). Bobwhites preferred dense herbaceous material and avoided bare ground and wooded area, whereas scaled quail selected for bare areas and woody cover and avoided herbaceous cover, in particular grass and grass/forb areas (Tables 6 and 7). During the nonbreeding season, bobwhites increased avoidance of bare ground and selected for dense herbaceous material (Table 7). Scaled quail increased their preference for woody cover during the nonbreeding season from 0.03 to Niche Breadth Indices for Cover Among Study Areas A simple measure of niche breadth was based on the total percent cover of grass, forb, grass/forb, and the amount of bare ground used by each species (Pianka 1973). Theoretically, the size of the niche breadth should be larger at the center of the breeding range than on the periphery of the distribution (Bilden 1965). However, both quails used the narrowest range of cover types at their respective core areas and a broader range on the peripheral/overlap areas. This seemingly contradictory statement can be best understood if the 2 core areas are viewed as sites with the largest amount of

44 30 Table 6. Selection indices for cover used by bobwhites and scaled quail on all 4 study areas combined, Texas, Selection indices Cover type Bobwhite Scaled Quail Grass 0.12 Forb 0.23 Grass/forb 0.27 Bare ground Woody

45 31 Table 7. Selection indices for cover used by bobwhite (BW) and scaled quail (SCQ) during the breeding and nonbreeding seasons on all 4 study areas combined, Texas, Selection indices by season Breeding Nonbreeding Cover type BW SCQ BW SCQ Grass Forb Grass/forb Bare ground Woody

46 32 species-specific "optimal" cover types. In the core areas, the life-history needs of the quail can be satisfied with a narrower range of cover types. In contrast, the coexisting populations on the peripheral study areas may have less "optimal" cover types to select from. Therefore, the quail may need a greater range of cover types to fulfill their needs. This view parallels the optimal foraging theory put forth by MacArthur and Pianka (1966) if cover type is substituted for food. According to this theory as the productivity of the environment decreases (e.g. towards the periphery of the range) the amount of search time increased and the itinerary is enlarged to include less suitable habitat. I suggest the concept of productivity in optimal foraging theory should be in relation to species-specific resources. For example, the Killam study area is probably less productive than Welder, in general, but probably more productive than Welder in terms of scaled quail resources. Niche Breadth Indices for Cover on All Study Areas Combined Niche breadth estimates for breeding and nonbreeding seasons were similar except for the bobwhites during the breeding season (Table 8). Apparently, bobwhites are less adaptable to arid conditions than are scaled quail (Schemnitz 1964). The drought conditions during the summer of 1980 were more unsuitable for bobwhite than scaled quail, and the former species may have used a broader range of cover types to satisfy their requirements. Another possibility is that the bobwhite is more of a generalist in that it can breed in a wide variety of cover types as demonstrated by its wide distribution.

47 33 Table 8. Niche breadth indices for cover (grass, forb and bare ground) used by bobwhites and scaled quail by each study area, on all 4 study areas combined, and by season, Texas, Niche breadth indices Grouping Bobwhite Scaled quail Welder 1.5 Chaparrosa Experiment Killam All study areas combined Breeding season Nonbreeding season

48 34 Food Analysis Annual Food Availability,Among Study Areas The percent frequency of available food items formed a gradient similar to the available cover (Table 9). The availability of food was different among the 4 study areas (Hotelling's T-square, P < 0.05). The major plant parts (seed, leaf, fruit, flower, bulb) and total animal matter of the 2 core areas were the most easily separated from the 4 sites (Appendix, Fig. C). The division of the major plant parts and animal matter into their finer categories improved the separation of all 4 areas (Appendix, Fig. D and Table The core area for bobwhites had the highest frequency of seeds (grass and forb), green leaves (grass and forb) and animal matter (Table 9 and Appendix, Table L). The frequency of animal matter was strongly correlated with the frequency of green leaves (r = 0.60, P < 0.05). The core area for scaled quail had the largest frequency of shrub seeds (34%), leaves (43%) and fruits (3%) (Appendix, Table The availability of food items on the peripheral sites of both quails were between these 2 extremes (Table 9). The major items that separated the Chaparrosa and Experiment Ranches were the larger percentage of forb seeds on the Chaparrosa (80% to 43%), and the higher frequency of snails on the Experiment Ranch (33% to 9%) (Appendix, Fig. D and Table L).